Interferon-a-induced G1 phase arrest through up-regulated ... - Nature

17 downloads 41 Views 2MB Size Report
G1/S transition was associated with diminuition of both ...... Dickinson, San Jose, CA) according to the manufacturer's instructions. .... Daniel Murozek helped us ...
Oncogene (1998) 16, 2075 ± 2086  1998 Stockton Press All rights reserved 0950 ± 9232/98 $12.00 http://www.stockton-press.co.uk/onc

Interferon-a-induced G1 phase arrest through up-regulated expression of CDK inhibitors, p19Ink4D and p21Cip1 in mouse macrophages Masaaki Matsuoka, Kenzaburo Tani and Shigetaka Asano Department of Hematology and Oncology, Institute of Medical Science, University of Tokyo, 4-6-1 Shirokanedai, Minato-ku, Tokyo 108, Japan

The mechanism of cell cycle arrest induced by interferon-a (IFN-a) was analysed using a mouse macrophage cell line, BAC1.2F5A. IFN-a added in media before mid-G1 prohibited cells from entering S phase. The blockage of G1/S transition was associated with diminuition of both cyclin D1/cdk4- and cyclin E/cdk2-associated kinase activities. G1 cyclin-associated kinase activities were down-regulated quickly after the addition of IFN-a. Cells treated with IFN-a contained excess amounts of cdk inhibitors which down-regulated G1 cyclin/cdk-associated kinase activities in the proliferating cells and this action was counteracted by exogenously-supplied recombinant cyclin D2/cdk4 complexes. In parallel, accumulation of p19Ink4D and p21Cip1, and their attachment to cdks were upregulated quickly after the addition of IFN-a. Expression of p19Ink4D and p21Cip1 was potentiated transcriptionally. We concluded that increased attachment of upregulated cdk inhibitors including p19Ink4D and p21Cip1 to G1 cyclin/ cdk complexes contributed to diminuition of G1 cyclin/ cdk-associated kinase activities and resulting G1 phase arrest during the early phase of treatment with IFN-a. Keywords: p19Ink4D; p21Cip1; interferon-a

Introduction Interferons (IFNs) are a family of secreted polypeptides which include three distinct classes, IFN-a, IFN-b and IFN-g with a variety of biological functions (Pestka et al., 1987). Their most important functions are the antiviral and the antiproliferative activities in mammalian cells, which have been utilized clinically for treatments of B- or C-type viral hepatitis and several neoplastic disorders. The post-receptor signal transduction pathways of IFNs as far as the transcriptional level have been intensively investigated over the past several years following the identi®cation of JAK kinases, IFNstimulated gene response elements (ISRE/GAS) and the transcription factors (ISGF-3/STATS) (Darnell et al., 1994; Ihle and Kerr, 1995). Two possible mechanisms for the antimitogenic activities of IFNs have been proposed; blockage of growth factor-stimulated pathways (Xu et al., 1995), and stimulation or inhibition of independent pathways such as those through JAK kinases and various transcription factors. G1 arrest is induced by IFNs as the ®nal common result of their antimitogenic activities in mammalian cells. The cell cycle progression is controlled by a family of cyclins and their speci®c catalytic partners, cyclinCorrespondence: M Matsuoka Received 12 August 1997; revised 21 November 1997; accepted 24 November 1997

dependent kinases or cdks (reviewed by Sherr, 1993). D-type cyclins (Lew et al., 1991; Matsushime et al., 1991) and cyclin E (Ko€ et al., 1991; Lew et al., 1991) govern the G1/S transition in association with their proper physiological partners, cdk4 (Matsushime et al., 1992) or cdk6 (Meyerson and Harlow, 1994), and cdk2 (Dulic et al., 1992; Ko€ et al., 1992), respectively. When cells enter G1 phase with mitogenic stimuli, the synthesis and the active complex formation of D-type cyclins and cdk4 are induced in early to middle G1 phase, and their complex formation and activities reach maximal levels at late G1 (Matsushime et al., 1992, 1994). Enforced overexpression of D-type cyclins shortened G1 phase in mammalian ®broblasts (Quelle et al., 1993). Reciprocally, the microinjection of anti-cyclin D antibodies caused G1 arrest, indicating that D-type cyclins are rate-limiting for G1/S transition (Quelle et al., 1993). The expression of cyclin E is induced later in G1 phase than D-type cyclins (Dulic et al., 1992; Ko€ et al., 1992) and is also responsible for the transition from G1 to S in mammalian cells (Ohtsubo and Roberts, 1993). For cells to complete the G1/S transition, the activities of both D-type cyclins and cyclin E are essential (Hatakeyama et al., 1994; Resnitzky and Reed, 1995; Resnitzky et al., 1994; van Den Heuvel and Harlow, 1993). In contrast to growth-promoting mitogenic signals, antimitogenic factors including transforming growth factor-beta or TGF-b (Polyak et al., 1994a), cAMP (Kato et al., 1994b), rapamycin (Nourse et al., 1994) and prostagrandin A2 (Gorospe et al., 1996) induced cell growth arrest by inhibiting cell cycle progression from G1 to S phase. The discovery of several cdk inhibitors has facilitated the elucidation of these mechanisms. Cdk inhibitors brake cell cycle progression by negatively and stoichiometrically regulating cyclin/cdk activities (Sherr and Roberts, 1995) and are the targets of various extracellular growth-modi®ers. Two families of mammalian cdk inhibitors have been identi®ed. One includes p21Cip1 (El-Deiry et al., 1993; Harper et al., 1993; Noda et al., 1994; Xiong et al., 1993), p27Kip1 (Polyak et al., 1994b; Toyoshima and Hunter, 1994) and p57Kip2 (Lee et al., 1995; Matsuoka et al., 1995). These inhibitors have similar N-terminal regions which are responsible for the blockage of cdk kinases. They can inhibit any G1 cyclin/cdk activity. The other family of cdk inhibitors includes INK4 cdk inhibitors (p16Ink4A, p15Ink4B, p18Ink4C and p19Ink4D) which inhibit cdk4 and cdk6 kinases but do not inhibit cdk2 activity (Chan et al., 1995; Guan et al., 1994; Hirai et al., 1995). p15, p18 and p19 are expressed ubiquitously while p16 is expressed in a limited range of normal tissues and in several cell lines. Overexpression of any one of these proteins results in G1 phase arrest of mammalian cells (Guan et al., 1994; Hirai et al., 1995; Quelle et al., 1995a).

Interferon a-induced G1 arrest M Matsuoka et al

2076

During normal cell cycle progression, p18 and p19 expression are dominantly induced after cells enter S and G2/M phase in a mouse macrophage cell line (Hirai et al., 1995) and the true function of p18 and p19 remains to be elucidated. The cDNA of another small protein p19ARF are generated by alternate splicing from the p16 gene. Overexpression of p19ARF induced cell cycle arrest (Quelle et al., 1995b). The genes of p16 or p19ARF and p15 exist contiguously on human chromosome 9p21 and are deleted or mutated in some malignancies (Hall and Peters, 1996; Kamb et al., 1994; Nobori et al., 1994), supporting the idea that cdk inhibitors can act as tumor suppressors. Cdk inhibitors also served as the common mediators of various biological processes including DNA-damages (Dulic et al., 1994), cell di€erentiation (Liu et al., 1996; Parker et al., 1995), cell to cell contact inhibition (Polyak et al., 1994a) and cell senescence (Hara et al., 1996; Noda et al., 1994). Recent studies indicated that oncogene signals converge to cyclin/cdk and cdk inhibitor (Lloyd et al., 1997). Several mechanisms have been suggested by a series of studies for TGF-b-induced inhibition of cell cycle progression, including the suppression of cdk4 synthesis for Mv1Lu mink lung epithelial cells (Ewen et al., 1993), inhibition of cyclin E/cdk2-associated kinases by p27Kip1 (Ko€ et al., 1993; Polyak et al., 1994a), and the increased expression of cdk inhibitors p15INK4b for human keratinocytes (Hannon and Beach, 1994). A comprehensive study (Reynisdottir et al., 1995) indicated that p27Kip1, p21Cip1 and p15Ink4B cooperated to induce cell cycle arrest in association with the subsequently occurring down-regulation of cdk4 levels. Expression of p15Ink4B mRNA and protein were induced primarily. Up-regulated p15 itself down-regulated cdk4-associated kinase activities while it competed with p27 for binding to cyclin D/cdk4. P27 released from cyclin D/cdk4 complexes attached to cdk2 and down-regulated cdk2-associated kinase activities (Reynisdottir and Massague, 1997). Down-regulation of cdc25A phosphatase by TGF-b was another cause reducing cdk4 and cdk6 activities in the p15-minus cell line (Lavarone and Massague, 1997). cAMP analogues and cAMP inducers inhibited cell growth and caused G1 phase arrest in mouse macrophages by increasing the accumulation of p27 and reducing the activities of CAK, cdk-activating kinase, indirectly (Kato et al., 1994b). Rapamycin induced G1 arrest by inhibiting elimination of p27 by IL-2 in T cells (Nourse et al., 1994). Prostaglandin A2 suppressed the expression of cyclin D and cdk4 and enhanced accumulation of p21 capable to inhibit cyclin E/cdk2 activities in human breast cancer cells (Gorospe et al., 1996). The relationship between IFN-g and cell cycle arrest has been characterized. In many cells, IFN-r upregulates p21Cip1 gene expression. Chin et al. (1996) demonstrated that IFN-r enhanced p21Cip1 gene expression through the IFN-r-Stat1 pathway. IFN-a induced cell cycle arrest by down-regulating the expression of cyclin D3 and cdc 25A in Daudi cells, Burkitt lymphoma cells (Tiefenbrun et al., 1996). In this study, we demonstrated that up-regulated accumulation of cdk inhibitors including p19 and p21 contributes to the induction of G1 phase arrest during early-phase IFN-a treatment in a mouse macrophage cell line.

Results Interferon-a blocked G1/S progression; the critical point was around mid-G1 BAC1.2F5A macrophages proliferated with about 18 ± 21 h of doubling time and were distributed randomly throughout the cell cycle in media containing CSF-1 (Figure 1a, left panel); 52.3% were within G1 phase, 41.2% within S phase and 6.5% within G2/M phase at 50% cell con¯uency. Preliminary experiments indicated that interferon (IFN)-a treatment blocked G1/S transition. To determine the optimal IFN-a concentration which blocked G1/S transition completely, the

a

b

c

Figure 1 Cell cycle arrest at mid G1 by IFN-a. (a) Exponentially growing BAC1.2F5A cells were incubated with or without IFN-a (2000 U/ml) for 20 h, and their cell cycle distribution was analysed by ¯ow cytometric analysis of DNA contents. (b) Cells were synchronized in early G1 phase by growth factor starvation for 18 h and stimulated with CSF-1 in the presence (lower panel) or absence (upper panel) of 2000 U/ml IFN-a. Cell cycle distribution of cells harvested at the indicated time points after stimulation with CSF-1 were analysed. (c) Synchronized cells were stimulated with CSF-1, and IFN-a (2000 U/ml) was added at indicated hours thereafter. Cells were harvested and cell cycle analyses were performed at 15 h after CSF-1 stimulation. Percentages of cells within S and G2/M phase were calculated with FACS analysis and compared to that of control cells treated with CSF-1 without IFN-a, which was adjusted as 100%

Interferon a-induced G1 arrest M Matsuoka et al

c IP Lysate

(–)

(–)

(+)

cAMP αE

(+)

Among D-type cyclins the most abundant is cyclin D1, and cdk4 is its major catalytic subunit in BAC1.2F5A cells (Kato et al., 1994b; Matsushime et al., 1994). Therefore, we analysed both cyclin D1- and cdk4associated kinase activities as representatives of D-type cyclin-dependent kinases. For screening experiments, we used the cells synchronized by CSF-1 starvation and restimulated with CSF-1 together with or without IFN-a for 12 h. When the lysates from cells stimulated with CSF-1 were immunoprecipitated with anti-cyclin D1 antibodies or anti-cdk4 antibodies, the precipitates contained pRb-kinase activities (Figure 2a and b, lane 2) as compared with those with nonimmune rabbit sera (Figure 2a and b, lane 1). The addition of IFN-a induced the marked reduction of kinase activities (Figure 2a and b, lane 3).

IFN αE

(–)

Inhibition of both cyclin D1/cdk4- and cyclin E/cdk2dependent kinase activities accompanied G1 arrest

IFN αE

(–)

treated with IFN-a. Similarily to cAMP or rapamycin treatment (Kato et al., 1994b), when added within 5 h, IFN-a prevented the majority of cells from entering S phase. When added thereafter, cells gradually became resistant to IFN-a treatment and entered S phase. IFNa added at more than 8 h after CSF-1 stimulation did not block G1/S transition e€ectively (Figure 1c). This suggested that the critical period for IFN-a was around the mid G1 when G1 cyclin-associated kinase activities were essential. We therefore focused on the activities of G1 cyclins, D-type cyclins and cyclin E.

IFN NRS

Lysate

IFN αD1

IP

IFN αD1

a

IFN NRS

proliferating cells were cultured in the presence of various concentrations of IFN-a for 18 h, then subjected to the cell cycle analysis. The percentage of cells within G1 phase increased gradually in proportion to the increase of added IFN-a (data not shown). More than 1000 units/ml of IFN-a inhibited G1/S progression almost completely. We used 2000 units/ml of IFN-a for the following experiments. When we cultured cells in the presence of 2000 units/ml of IFN-a for 20 h, 88.6% of cells distributed in G1 phase (Figure 1a, right panel), indicating that IFN-a blocked G1 phase progression. Macrophages deprived of CSF-1 for 18 h were arrested in early G1 phase (Tushinski and Stanley, 1985). They were restimulated with CSF-1 in the absence or the presence of 2000 U/ml IFN-a and synchronized cell cycle progression was monitored at various time points (Figure 1b). Normal G1/S transition began to take place after 10 ± 12 h and the majority of cells were in S phase after 15 h (Figure 1b, upper panel). In the presence of IFN-a, however, almost all cells remained in G1 phase even 15 h after stimulation with CSF-1, con®rming that IFN-a induced G1 phase arrest (Figure 1b, lower panel). IFN-a was added at various time points after CSF-1 stimulation to determine which period in G1 phase was critical for induction of G1 arrest with IFN-a. In this study, we calculated percentages of cells within both S and G2/M 15 h after stimulation with CSF-1 by FACScan ¯ow cytometrical analysis of DNA content to determine the percentages of cells entering S phase, and compared these values with the values in cells not

pRb

pRb

1 2

(+)

IFN αCDK2

(–)

Lysate

IFN αCDK2

IFN αCDK4

(–)

IP

IFN NRS

IFN αCDK4

Lysate

IFN NRS

IP

4

3 d

b

3

(–)

(–)

(+)

cAMP αCDK2

1

2

pRb

pRb

1

2

3 1

2

3

4

Figure 2 E€ects of IFN-a on G1 cyclin- and cdk-associated kinase activities. (a, b, c and d) BAC1.2F5A cells deprived of CSF-1 for 18 h were stimulated with CSF-1 alone ((7)-IFN-a, lanes 1 and 2), or together with IFN-a((+)-IFN-a, lane 3) or 8 Br-cAMP (cAMP, lane 4). After 12 h, cells were harvested and the lysates were immunoprecipitated with nonimmune rabbit sera (NRS), and antibodies to cyclin D1 (a D1), cdk4 (a cdk4), cyclin E (a E) or cdk2 (acdk2) and assayed for pRb kinase activities. For pRb substrates, recombinant GST-fused Rb large pocket protein (a and c) or recombinant GST-fused Rb C-terminal protein (b and d) were used

2077

Interferon a-induced G1 arrest M Matsuoka et al

2078

Another G1 cyclin, cyclin E, and its major catalytic partner cdk2 work at G1 phase, slightly later than Dtype cyclins. The immunoprecipitates of lysates from macrophages stimulated with CSF-1 with antibodies to cyclin E and to cdk2 contained Rb kinase activities (Figure 2c and d, lane 2). The anti-cdk2 immunoprecipitates of cell lysates from mammalian cells around the G1/S boundary contained cyclin E/cdk2-associated kinases as well as cyclin A/cdk2-associated kinase activities (Dulic et al., 1992; Ko€ et al., 1992). Addition of IFN-a reduced both cyclin E- and cdk2associated kinase activities markedly (Figure 2c and d, lane 3). As another control, cell lysates treated with the cAMP analogue, 8 Br-cAMP were also analysed for cyclin E- or cdk2-associated kinase activities. As expected from a previous study (Kato et al., 1994b), cAMP down-regulated cyclin E- or cdk2-associated kinase activities (Figure 2c and d, lane 4). For cells to complete the progression from G1 to S phase, both cyclin D/cdk4- and cyclin E/cdk2-associated kinase activities, targets of which are supposed to be distinct from each other, are necessary (Hatakeyama et al., 1994; Resnitzky and Reed, 1995). Therefore, the reduction of both cyclin D1/cdk4- and cyclin E/cdk2associated kinase activities cooperated to induce G1 phase arrest. IFN-a reduced G1 cyclin/cdk activities quickly To determine how quick IFN-a down-regulated G1 cyclin/cdk activities, asynchronously-growing subcon¯uent BAC12F5A cells were incubated with new media containing fresh CSF-1 together with or without IFN-a. After the splitting of con¯uent BAC12F5A cells by a one ®fteenth ratio, it took 3 or 4 days for us to acquire 80% cell con¯uence, in which state 60% of cells were in the G1 phase (Figure 3c). When we started to incubate the cells with IFN-a, we replaced the old media with small volumes of new media containing fresh CSF-1 with IFNa. To be precise, we needed cells which were fed with fresh CSF-1 without IFN-a as controls because the expression of various cell cycle components and G1 cyclin/cdks-associated kinase activities were minimally a€ected by the fresh CSF-1 in the new media (see below). The addition of only fresh CSF-1 gradually reduced the percentage of cells within G1 phase and reciprocally increased cells within S and G2/M phase by potentiating G1 phase progression (Figure 3c, lanes 1, 2, 4, 6). In the presence of IFN-a together with fresh CSF-1, however, the percentage of cells within G1 phase increased gradually after the addition of IFN-a, re¯ecting blockage of G1/S transition by IFN-a (lanes 3, 5, 7). The variation of cdk4- or cdk2-associated Rb kinase activities in the asynchronously-growing cells during treatment with IFN-a was determined with this system (Figure 3a and b). Re¯ecting mitogenic stimulus of fresh CSF-1 in new media, cdk4-associated kinase activities increased to one and a half fold the original level at 6 h in the control cells (Figure 3a, lane 4). Cdk2-associated kinase activities were not in¯uenced by media change (Figure 3b). In the presence of IFN-a, cdk4- or cdk2associated kinase activities fell to less than one-®fth the control levels within 3 h after addition of IFN-a (Figure 3a and b, lanes for IFN-a+). The rapidity of IFN-a's e€ect on G1 cyclin activities was in contrast to that of the other antimitogenic agents including TGF-b, which

needs 8 h or more to down-regulate cdk2-associated kinase activities to half the normal levels (Reynisdottir et al., 1995). We focused on the mechanisms downregulating G1-cyclin activities during the early phase of IFN-a treatment within 9 h. Cells incubated with IFN-a contained inhibitors which were able to reduce G1 cyclin/cdk-associated kinase activities in the proliferating cells One possible cause of the reduction in cdk4- or cdk2associated Rb kinase activities was supposed to be upregulation of cdk inhibitors which act stoichiometrically and are heat-stable (Sherr and Roberts, 1995). To test this possibility, we performed mixing experiments to determine whether the lysates from cells incubated with IFN-a contained excess amounts of cdk inhibitors which could down-regulate the G1 cyclin/cdk-associated kinase activities in the lysates from the proliferating cells not treated with IFN-a. Immune complex cdk2-associated kinase assays were performed under the condition of linear correlation between the amounts of added cell lysates and immune complex kinase activities (Figure 4a). Nearly equal amounts of lysates from cells incubated with IFN-a for 9 h could inhibit cdk2 kinase activities in the lysates from control cells (Figure 4a). Similar experiments assessing the activities inhibiting cdk4-associated kinase gave similar results (Figure 4b), indicating that there were sucient cdk inhibitors in the lysates from cells treated with IFN-a to down-regulate not only cdk2- but cdk4associated kinase activities in the equivalent amounts of normal cell lysates.

a IFN-α

P –

1 –

2 –

3 +

4 –

5 +

6 –

7 +

Time (hr)

0

0

3

3

6

6

9

9

IP: α cdk4

pRb

b

IP: α cdk2

pRb

c % GI %S % G2/M

60 30 10

60 29 11

59 30 11

56 39 5

63 30 7

45 44 11

77 15 8

Figure 3 IFN-a down-regulated both cdk4- and cdk2-associated kinase activities quickly. Asynchronously-growing subcon¯uent BAC1.2F5A cells were cultured in new media containing fresh CSF-1 with or without IFN-a (indicated + or 7) for the indicated periods. Cell lysates were assayed for immune complex cdk4- (a) or cdk2- (b) associated kinase activities with substrates of pRb and 16105 cells were analysed for cell cycle distribution (c). Lane P indicates the lane for immunoprecipitation with preimmune rabbit sera

Interferon a-induced G1 arrest M Matsuoka et al

a P IFN (–) +

Lysate Cyclin E

2079

IP: α cyclin E 1 2 3 IFN IFN — (–) (–) – + +

4 IFN (+) +

pRb b (Ig) cdk2

Figure 4 The cells treated with IFN-a contained excess inhibitors which down-regulated cdk2- or cdk4-associated kinase activities. The asynchronously-growing subcon¯uent cells were co-cultured with new media containing fresh CSF-1 with or without IFN-a for 9 h, subsequently harvested and lysed in the H1 kinase bu€er containing protease and phosphatase inhibitors. Indicated amounts of cleared supernatants from the two types of cell (IFN-a, 7 or +) were mixed and incubated for 30 min at 238C before immune complexes cdk2- (a) or cdk4- (b) associated kinase reaction was performed with pRb proteins as substrates. The ®nal volume of each reaction mixture was adjusted with HI kinase bu€er. 32P radioisotope counts of phosphorylated pRb were monitored with Bioimage analyzer BAS-2000II (Fuji Photo Film Co, Ltd, Shizuoka, Japan). The presented results are representative of three independent experiments

Exogenously-added recombinant cyclin D2/cdk4 counteracted inhibitory activities of the lysates from cells treated with IFN-a Human cyclin E generated baculovirally in the insect cells was mixed and incubated with cell lysates for 30 min. Immune complex human cyclin E-associated kinase activities were determined with these lysates as described (Polyak et al., 1994a). Human cyclin E bound to endogenous mouse cdk2 derived from the control cell lysates gave rise to kinase activities (Figure 5a, lane 3) while neither single cyclin E or single cell lysate, nor the immunoprecipitates with preimmune rabbit sera did (lane P, 1, 2). We could get only one®fth the kinase activities of the controls from cells treated with IFN-a for 9 h (lane 4). To examine the complex formation between human cyclin E and endogenous mouse cdk2, the mixed cell lysates were immunoprecipitated with antibodies to human cyclin E and the precipitates were electrophoresed on SDS ± PAGE gel, then immunoblotted with antibodies to cdk2 (Figure 5b). The alternatively-spliced larger form of cdk2, 42 kd cdk2 was expressed in the mouse cells in addition to the usual form, 34 kd cdk2 (Kato and Sherr, 1993). In both the control cells and cells treated with IFN-a, similar levels of human cyclin E/cdk2 complexes existed (lanes 3, 4). Thus, we con®rmed that the IFN-a-treated cell lysates contained activities negatively modifying cyclin E/cdk2 complexes. Based on the hypothesis that stoichiometricallyacting cdk inhibitors down-regulated kinase activities, which could be neutralized and trapped by excess amounts of cyclin/cdk complexes (Polyak et al., 1994a), we added the baculovirally-produced cyclin D2/cdk4 in addition to human cyclin E to the above lysates before assays of human cyclin E-associated kinase activities

Figure 5 The kinase activities derived from exogenously-added human cyclin E complexed with endogenous mouse cdk2 from BAC1.2F5A cells were suppressed by excess inhibitors residing in the cells treated with IFN-a. Insect cell lysates containing or not containing human cyclin E generated baculovirally were mixed and incubated for 30 min at 238C with or without 150 mg of the cell lysates in the H1 kinase bu€er from asynchronously-growing cells treated with or without IFN-a (IFN-a, + or 7) for 9 h, then immune complex human cyclin E-associated kinase assays were conducted (a) or the mixed lysates were precipitated with antibodies to human cyclin E or preimmune rabbit sera (indicated as P) and the precipitates were immunoblotted with antibodies to cdk2 (b). The amounts of added human cyclin E were estimated to be about threefold the physiological level of cyclin E

a Insect



IFN-α



D2

/k4





+

D2

K4

+

+

D2/

k4

+

HI

b D2/cdk4 (µg) IFN-α

0

0

0.1

0.3

1.0



+

+

+

+

HI

Figure 6 Excess recombinant cyclin D2/cdk4 complexes counteracted the inhibitory action in the IFN-a-treated cells. (a) Insect cell lysates containing human cyclin E were mixed and incubated for 30 min at 238C with 150 mg of the lysates from asynchronously-growing cells treated with or without IFN-a (IFN-a, + or 7) for 9 h in the presence of the insect cell lysates not infected (indicated as 7) or those containing cyclin D2, cdk4 or cyclin D2/cdk4, after which immune complex human cyclin E-associated kinase assays were conducted with histone H1 as substrate. The added amounts of cyclin D2, cdk4 or cyclin D2/cdk4 were estimated as about threefold the physiological levels. (b) Insect cell lysates containing human cyclin E were mixed and incubated for 30 min at 238C with 150 mg of the lysates from asynchronously-growing cells treated with or without IFN-a (IFN-a, + or 7) for 9 h in the presence of the insect cell lysates containing indicated amounts of cyclin D2/cdk4, then immune complex human cyclin E-associated kinase assays were conducted with histone H1 as substrate

Interferon a-induced G1 arrest M Matsuoka et al

2080

(Figure 6a). Cyclin D2/cdk4 but not single cyclin D2 nor single cdk4 restored the kinase activities to the control level. This was because cyclin D2/cdk4 complexes trapped cdk inhibitors more eciently than cyclin E/cdk2 complexes or single cyclin D2 or single cdk4. Furthermore, cyclin D2/cdk4 complexes restored cyclin E activities in a dose-dependent manner (Figure 6b). All these results indicated that cells treated with IFN-a contained excess cdk inhibitors which were neutralized when enough cyclin D2/cdk4 was supplied exogenously. Thus we concluded that up-regulated accumulation of cdk inhibitors contributed to the decrease of G1 cyclin activities in the cells treated wtih IFN-a. The protein levels of p19 and p21 were quickly upregulated after addition of IFN-a To see which cdk inhibitors are responsible for diminuition of G1 cyclin activities, we determined temporal pro®les of known cdk inhibitor levels in contrast to those of cyclin levels as positive regulators and cdk levels during the early phase of treatment with IFN-a (Figure 7). Immunoblot analyses were performed with the antibodies to cdk inhibitors including p19Ink4D, p21Cip1 and p27Kip1 and those to cyclin D1 and cyclin E using whole cell lysates from cells before treatment with IFN-a (indicated as 0) and those cultured in new media with fresh CSF-1 together with or without IFN-a (IFN-a, + or 7) for the indicated times (3 ± 9 h) (Figure 7a). We again made the lysates

a

b

IFN-α Time (hr.)

p19 p21

Lysate 1 2 3 4 5 6 7 – – + – + – + 0 3 3 6 6 9 9

IFN-α Time (hr.)

for the control lanes from cells shifted to the new media containing only fresh CSF-1 (IFN-a, 7) because fresh CSF-1 slightly changed cell cycle distribution (Figure 3c) and the expression of various cell cycle components including cyclins, cdks and cdk inhibitors. We did not conduct the same experiments with antibodies to p15Ink4A or p15Ink4B as the genes for these proteins were deleted in this cell line (Quelle et al., 1995a). The levels of total p19 and p21 were up-regulated fourfold the control within 3 h (Figure 7a, p19 and p21, lanes 1, 3), and remained elevated at 6 ± 9 h (lanes 5, 7) after the addition of IFN-a. Fresh CSF-1 without IFN-a increased the relative amounts of total p19 and p21 (Figure 7a, p19 and p21, lanes 1, 2, 4, 6) possibly because it increased the relative percentage of cells within S and G2/M phases, and such cells contain relatively larger amounts of p19 and p21 (Figure 3c) (Hirai et al., 1995; Li et al., 1944). The levels of total p27 decreased to about one third the original level by replacement of the old media with new media containing fresh CSF-1, irrespective of IFN-a (Figure 7a, p27). The levels of total cyclin D1 and cyclin E were basically invariant during treatment with IFN-a (Figure 7a D1 and E). To determine the amount of the various cdk inhibitors and cyclins attached to cdks during treatment with IFN-a, sequential immunoprecipitation-immunoblotting was performed. The same series of cell lysates used in the kinase assays described above (Figure 3) were immunoprecipitated with the anti-

IP: α cdk4 P 1 2 3 4 5 6 7 – – – + – + – + 0 0 3 3 6 6 9 9

c IFN-α Time (hr.) (Ig)

IP: α cdk2 P 1 2 3 4 5 6 7 – – – + – + – + 0 0 3 3 6 6 9 9

cdk2

cdk4

(Ig) D1 E

p27 p19

p21 D1

p21 cdk2

E

(Ig)

p27

p27 Figure 7 IFN-a treatment up-regulated the levels of p19 and p21 quickly. Asynchronously-growing subcon¯uent BAC1.2F5A cells were cultured in new media containing fresh CSF-1 with or without IFN-a (indicated + or 7) for the indicated periods. (a) The cell lysates (100 mg) prepared in NP40 lysis bu€er were immunoblotted with antibodies to p19, p21, p27, cyclin D1 and cyclin E. (b) The cell lysates (400 mg) prepared in EBC bu€er were precipitated with antibodies to cdk4 and the precipitates were immunoblotted with indicated antibodies. The lane p indicates the lanes for immunoprecipitation with preimmune rabbit sera. (c) The cell lysates (200 mg) prepared in Tween 20 lysis bu€er were precipitated with antibodies to cdk2 and the precipitates were sequentially immunoblotted with indicated antibodies. The blotting with anti-p27 antibodies was performed with the un-erased membrane previously blotted with anti-cdk2 antibodies

Interferon a-induced G1 arrest M Matsuoka et al

bodies to cdk4 (Figure 7b) or cdk2 (Figure 7c), and the precipitates were electrophoresed, fractionated on the SDS ± PAGE gels and immunoblotted with antibodies to cdk2, cdk4 or a series of antibodies to various proteins bound to cdk4 or cdk2 (Figure 7b and c). The protein levels of total cdk4 and cyclin D1 bound to cdk4 were constant after new media with fresh CSF1 were added irrespective of IFN-a (Figure 7b, cdk4 and D1). The levels of p19 bound to cdk4 were gradually and slightly up-regulated by fresh CSF-1 without IFN-a (Figure 7b, p19, lanes 1, 2, 4, 6) again because fresh CSF-1 transiently potentiated G1 phase progression and increased the non-G1 cell population. In the presence of IFN-a, they were up-regulated sixfold the original level within 3 h (Figure 7b, p19, lane 3), and remained elevated at 6 ± 9 h (lanes 5, 7). The levels of p21 bound to cdk4 were apparently upregulated at 3 h by treatment with IFN-a (Figure 7b, cdk4, D1 and p21, lanes 2, 3), but decreased thereafter in proportion to the levels of cyclin D1/cdk4 complexes. The levels of p27 bound to cdk4 gradually increased after the addition of fresh CSF-1, irrespective of IFN-a (Figure 7b, p27). Note that up-regulated p19 did not seem to compete with p27 for binding to cdk4 during treatment with IFN-a. Up-regulated p15 did compete with p27 for binding to cdk4 during treatment with TGF-b (Reynisdottir et al., 1995; Reynisdottir and Massague, 1996). The levels of total cdk2 and cyclin E bound to cdk2 were not a€ected by replacement of media (Figure 7c, cdk2 and cyclin E). Up-regulation of p21 bound to cdk2 by IFN-a was observed at 3 h (Figure 7c, p21, lanes 1, 2, 3). The levels increased ®vefold the control levels at 6 ± 9 h (lanes 4, 5, 6, 7). Note that p21 levels decreased to two thirds or less the original level after transfer of cells to new media containing only fresh CSF-1 (lanes 1, 2, 4, 6). p21 levels were always higher during treatment with IFN-a than their original level (lanes 1, 3, 5, 7). p27 attached to cdk2 was upregulated about threefold the control level at 9 h (Figure 7c, p27, lanes 6 and 7) by treatment with IFNa. The levels of p27 bound to cdk2 were reduced to a half or less the original level only by replacement of media with those containing fresh CSF-1 (Figure 7c, p27, lanes 1, 2, 4, 6). Therefore, up-regulation of p27 bound to cdk2 by IFN-a was relatively mild compared with the original level. Collectively, up-regulation of total p19 and p19 bound to cdk4, and minimally that of total p21 and p21 bound to cdk2 is supposed to be responsible for the down-regulation of cyclin D1/cdk4-associated kinase activities during early phase treatment of IFNa. We could also conclude that upregulated p21 participated in down-regulation of cyclin E/cdk2associated kinase activities. p19 levels were up-regulated by IFN-a in a doseresponsive fashion We examined variation of the p19 protein levels in the lysates from asynchronously-growing cells treated with several di€erent concentrations of IFN-a for 12 h to see whether or not they increased in parallel to the concentration of IFN-a (Figure 8a). The internal controls were the levels of cyclin D1 proteins (Figure 8b). The p19 levels were up-regulated about twofold in

the presence of 1000 U/ml IFN-a (Figure 8a, lanes 1, 2), and about four- or ®vefold in the presence of 2000 U/ml or 4000 U/ml IFN-a (lanes 3 and 4). However, the levels of cyclin D1 protein were almost constant irrespective of IFN-a concentration (Figure 8b). This observation indicated that gene expression of p19 was induced by IFN-a in a dose-responsive fashion in this cell line. IFN-a up-regulated mRNA levels of p19 and p21 Northern blot analysis was performed to see the action point of IFN-a on gene expression of p19 and p21. p19 and p21 mRNA levels increased at 3 h, 6 h and 9 h or 12 h after addition of IFN-a (Figure 9a and b). Upregulation of steady-state p19 and p21 mRNA levels by IFN-a is about three times the controls and seems to be a little smaller than that of p19 and p21 protein levels by IFN-a (Figure 7a). We speculate that the stability of those proteins also increased during treatment with IFN-a. The activities of cdk2 CAK and the protein levels of cdc25A were not down-regulated by IFN-a With the results shown above, we could not completely exclude the possibility that other post-translational modi®cations contributed to down-regulation of G1 cyclin activities. An important post-translational modi®cation of cyclin/cdk is mediated by the cdkactivating kinases, CAKs, which phosphorylate cyclinbound cdks on a single threonine residue (Thr 160 in cdk2 and Thr 172 in cdk4) (Fisher and Morgan, 1994; Kato et al., 1994a; Matsuoka et al., 1994; Morgan, 1995). These modi®cations are essential for cyclin-cdks to become active holoenzymes and might themselves be the targets of antiproliferative factors. With the lysates from synchronized cells cultured with or without IFNa or cAMP for 12 h, we performed immune complex cdk2 CAK assays (Figure 10a) (Matsuoka et al., 1994). The substrates were the bacterially-generated Histidine-

a

1

2

3

4

(kd) 26

p19 14

b

48 D1 32

Figure 8 p19 levels were up-regulated during treatment with IFN-a in a dose-dependent manner. Subcon¯uent cells were cocultured with new media containing fresh CSF-1 without or with various concentrations of IFN-a (lane 1-0 U/ml, lane 2-1000 U/ ml, lane 3-2000 U/ml, lane 4-4000 U/ml) for 12 h. Cell lysates (100 mg) in NP40 lysis bu€er were immunoblotted with antibodies to p19 (a) and cyclin D1 (b)

2081

Interferon a-induced G1 arrest M Matsuoka et al

1 3 –

2 3 +

3 6 –

4 6 +

5 12 –

6 12 +

Among the various biological activities of IFNs, antimitogenic activity has attracted a great deal of attention and has been utilized for treatment of neoplasms. The basic mechanism by which IFNs block cell cycle progression is now being elucidated. Murine BAC1.2F5A is a macrophage cell line suitable for analysis of the e€ects of IFNs on the cell cycle because cells proliferate only in the presence of CSF-1, are easily synchronized at early G1 phase by deprivation of CSF-1, and are arrested in G1 phase with IFN-a treatment. CSF-1 is required until cells reach the late G1 restriction point, after which they progress through the remainder of the cell cycle automatically. Addition of IFN-a prior to the mid G1 phase prevented cells from passing the G1/S boundary (Figure 1c). We therefore focused on the activities of G1 cyclins, D-type cyclins/cdk4 and cyclin E/cdk2. Using several antibodies, immune complex kinase assays were performed. As shown above, in asynchronously-growing and synchronized cells treated

a p19 IP

Lysate

cAMP α CDK7

Time (hr.) INF-α

Discussion

(+) IFN α CDK7

a

tion of cdc25A activities was the primary cause for cdk2-associated kinase activities in the mouse macrophage cell lines without determining phosphatase activities of cdc25A.

(–) IFN α CDK7

tagged cdk2 and the antibodies were against cdk7 (Poon et al., 1993). Cdk2 CAK activities were not a€ected by the presence of IFN-a or cAMP. We obtained similar results with the lysates from asynchronously-growing cells treated or not treated with IFN-a for 9 h (data not shown). Another possible candidate, the activities of which might be down-regulated during treatment with IFN-a, was cdc25A, which was found to regulate G1/S progression of mammalian cells (Gu et al., 1992; Jinno et al., 1994; Terada et al., 1995). The expression of cdc25A was down-regulated by IFN-a in Daudi cells (Tieferbrun et al., 1996). Therefore, the protein levels of double 65 and 80 kd mouse cdc25A in BAC1.2F5 cells were assessed with immunoprecipitation and immunoblotting (Figure 10b). To assure that signals really indicated cdc25A, in the last lane the immunizing peptides were added at the immunoprecipitation step to cancel out the signals (Figure 10b, lane 8). Irrespective of IFN-a, the protein levels of cdc25A were constant for at least 12 h after the addition of IFN-a in the asynchronously-growing cells. Although unlikely from above experiments, we could not completely rule out the possibility that down-regula-

(–) IFN NRS

2082

β-actin CDK2

1 b Time (hr.) INF-α

1 3 –

2 3 +

3 6 –

4 6 +

5 9 –

6 9 +

p21

b 1 peptide – INF-α – Time (hr.) 0 Blot: cdc25A

EF-1 α

Figure 9 IFN-a increased mRNA levels of p19 and p21. Asynchronously-growing subcon¯uent BAC1.2F5A cells were cultured in new media containing fresh CSF-1 with or without IFN-a (indicated + or 7) for the indicated periods. Fractionated mRNAs transferred onto nitrocellulose ®lters were hybridized with a-32P-dCTP-labeled cDNA of p19 (a) or p21 (b). The same nitrocellulose ®lters were rehybridized with a-32P-dCTP-labeled human b-actin (a) or human EF-1a (b) to indicate the applied amounts of mRNA

2

3

4

IP: α cdc25A 2 3 4 5 6 7 – – – – – – – + – + – + 6 6 9 9 12 12

8 + – 0 — 84 kd — 62 kd

Figure 10 Neither the cdk2 CAK activities nor the protein levels of cdc25A were down-regulated during treatment with IFN-a. (a) Subcon¯uent BAC1.2F5A cells were synchronized at early G1 by starvation of CSF-1 and restimulated with media containing fresh CSF-1 with or without IFN-a or with cAMP ((7)-IFN, (+)-IFN or (+)-cAMP) for 12 h. Cell lysates (200 mg) in the H1 kinase bu€er were immunoprecipitated with the non-immune rabbit sera or the antibodies to cdk7 and the precipitates were assayed for cdk2 CAK activities with 3 mg of bacterially-produced His-tagged cdk2 as substrate. (b) Asynchronously-growing subcon¯uent BAC1.2F5A cells were cultured in new media containing fresh CSF-1 with or without IFN-a (indicated + or 7) for the indicated periods. Cell lysates in Tween 20 lysis bu€er were precipitated with antibodies to cdc25A and the precipitates were immunoblotted with the same antibodies. The immunizing peptides (1 mg) were added to the lysates for the last lane

Interferon a-induced G1 arrest M Matsuoka et al

with CSF-1 plus IFN-a, both cyclin E/cdk2 and cyclin D1/cdk4 activities were reduced markedly and very quickly (Figures 2 and 3). The temporal pro®le of G1cyclin-associated activities during treatment with IFN-a indicated that IFN-a down-regulated them within 3 h. Therefore searches for mechanisms down-regulating cyclins/cdks activities during early-phase IFN-a treatment were performed. A possible mechanism blocking the activities of G1 cyclin D1/cdk was the increased attachment of various cdk inhibitors to cdks which act stoichiometrically and are heat-stable (Sherr and Roberts, 1995). Screening kinase assays using boiled lysates as the source of inhibitors indicated that cells treated with IFN-a contained more heat-stable cdk2 inhibitors than control cells (data not shown). Known cdk inhibitors act stoichiometrically and should be abundant enough to down-regulate G1 cyclin/cdk activities. We examined whether or not excess cdk inhibitors existed in the cells treated with IFN-a. The mixing experiments revealed that the lysates from cells treated with IFNa contained enough inhibitory activities to downregulate G1 cyclin/cdk-associated kinase contained in the equivalent amounts of lysates from control cells (Figure 4). Furthermore, sucient amounts of exogenously-added baculovirally-generated cyclin D2/ cdk4 complexes counteracted and neutralized these inhibitory activities residing in IFN-a-treated cell lysates (Figures 5 and 6). All these observations substantiated the idea that cells treated with IFN-a contained excess amounts of cdk inhibitors which down-regulated G1 cyclin-associated kinase activities stoichiometrically. To assess the actual mediators of the IFN-a e€ect on the cell cycle, we next determined time courses of the levels of known cdk inhibitors as well as the levels of cdk inhibitors bound to cdks in the cells during treatment with IFN-a. The levels of total p19 or p21 and the levels of p19 or p21 bound to cdks were quickly up-regulated during treatment with IFN-a in parallel to down-regulation of G1 cyclin activities (Figures 3 and 7), indicating that up-regulated p19 or p21 could down-regulate cyclin D/cdk4- or cyclin E/ cdk2-associated kinase activities, respectively, during the early phase of IFN-a treatment. mRNA levels of p19 and p21 were upregulated similarily during treatment with IFN-a (Figure 9) while increase of protein stability was also supposed to contribute to upregulation of these protein levels. Although the precise function of p19 still remains to be clari®ed, it is natural for us to think that upregulated accumulation of p19 reduced cyclin D1/cdk4 activities during treatment with IFN-a. The previous experiments indicated that recombinant p19 selectively inhibited the activities of D type cyclins/cdk in vitro in a dose-responsive fashion and enforced overexpression of p19 induced G1 phase arrest as well as the disappearance of cyclin D kinase activities in the mammalian cells (Hirai et al., 1995). Currently, we cannot completely rule out the possibility that other cdk inhibitors such as p18INK4C or p57KIP2 participate in the inhibition of G1 cyclin/cdk-associated kinase activities during treatment with IFN-a. Northern blot analysis had indicated that the expression of p18 mRNA oscillated like that of p19 mRNA during normal cell cycle progression of BAC1.2F5 cells (Hirai

et al., 1995). During treatment with IFN-a, however, the expression of p18 mRNA was not enhanced (M Matsuoka and S Asano, unpublished observation) while that of p19 mRNA was potentiated (Figure 9a), suggesting that p18 does not seem to behave like p19 during treatment with IFN-a. Regulation of cell cycle components by IFN-a was distinct from that by TGF-b. The mechanism of TGFb-mediated G1 phase arrest has been extensively analysed (Reynisdottir et al., 1995; Reynisdottir and Massague, 1997; Lavarone and Massague, 1997). p27Kip1, p21Cip1 and p15Ink4B cooperated to induce cell cycle arrest. Expression of p15Ink4B mRNA and protein were induced primarily. Up-regulated p15Ink4B itself down-regulated cdk4-associated kinase activities while it competed with p27 for binding to cyclin D/cdk4. P27 released from cyclin D/cdk4 complexes attached to cdk2 and down-regulated cdk2-associated kinase activities. IFN-a similarily up-regulated p19, a family member of Ink4 type cyclin inhibitors. However, upregulated p19 did not seem to compete with p27 for binding to cyclin D/cdk4. Instead, quickly up-regulated levels of total p21 and p21 bound to cdk2 reduced cdk2-associated kinase activities during treatment with IFN-a. It remains to be clari®ed whether other posttranslational modi®cations than cdk inhibitors contributes to down-regulation of G1 cyclin/cdks activities during treatment with IFN-a. Two types of positive cdk modi®ers have been well characterized, one of which is mediated by the cdk-activating kinases, CAKs (Fisher and Morgan, 1994; Kato et al., 1994a; Matsuoka et al., 1994; Morgan, 1995). CAKs might themselves be the targets of antiproliferative factors (Kato et al., 1994b). The phosphatase cdc25A functions as a positive regulator of G1/S transition (Jinno et al., 1994) by potentiating G1 cyclin/cdk activities by dephosphorylating tyrosine 15 of cdk2 or tyrosine 17 of cdk4 (Gu et al., 1992; Terada et al., 1995), and may be another target of antimitogenic signals. In Daudi cells, the mRNA and protein levels of cdc25A were quickly down-regulated by treatment with IFN-a (Tiefenbrun et al., 1996). In MCF7 cells in which p15 gene was deleted but which were still sensitive to antiproliferative action of TGF-b, downregulation of cdc25A phosphatase gene expression by TGF-b contributed to inactivation of cdk4 and cdk6 function (Lavarone and Massague, 1997). In BAC12F5 cells treated with IFN-a, however, the cdk2 CAK activities were intact and the protein levels of cdc25A were constant, suggesting that neither cdk2 CAK nor cdc25A are the mediators of IFN-a during early-phase treatment (Figure 9). To be exact, we need to determine the dephosphorylating activities of cdc25A during treatment with IFN-a to completely rule out a role for cdc25A in IFN-a-induced G1 phase arrest. Furthermore, we did not exclude the possibility that another unknown post-translational modi®cation was responsible for the IFN-a-mediated down-regulation of G1 cyclin/cdk-associated kinase activities especially in the later phase of IFN-a treatment. We have few similar evidences supporting the generality of the phenomenon that p19 or p21 is upregulated during treatment with IFN-a. It was demonstrated that p21 levels were up-regulated in a prostatic cancer cell line and Daudi Burkitt's

2083

Interferon a-induced G1 arrest M Matsuoka et al

2084

lymphoma cells during treatment with IFN-a (Hobeika et al., 1997). The gene expression of p19 was not analysed there. We ourselves did not observe that IFNa up-regulated the expression of p21 nor p19 gene in Daudi cells (data not shown). In 32D cells, mouse myeloblastic cells which are partially sensitive to IFNa, p19 and p21 protein levels were only minimally upregulated (data not shown). We speculate that p19 and p21 gene expression is induced by IFN-a especially in myelomonocytoid cells. In conclusion, down-regulation of cyclin D1/cdk4and cyclin E/cdk2-associated kinase activities and resulting G1 phase arrest during treatment with IFNa were at least partially mediated by quick upregulation of cdk inhibitors including p19 and p21 in macrophage cells. Materials and methods Cell culture and cell cycle analyses Murine BAC1.2F5A macrophages were grown in Dulbecco's modi®ed Eagle's medium (DMEM) supplemented with 20% fetal bovine serum (FBS), 100 U of penicillin G per ml and 100 mg of streptomycin per ml and 25% L cellconditioned medium, as described (Matsushime et al., 1991). These cells are dependent on CSF-1 for proliferation and survival. Macrophages were arrested in early G1 by deprivation of CSF-1 for 18 h and stimulated with CSF1 to reenter the cell cycle in a synchronized fashion. About 80% con¯uent (subcon¯uent) cells were used both for IFNa treatment of asynchronously-growing cells and for CSF-1 deprivation followed by restimulation with CSF-1 together with IFN-a of synchronous cells unless otherwise cited. Usually, 3 ml of media with CSF-1 with or without 2000 U/ml IFN-a was added to 10 cm dishes. IFN-a was kindly supplied by Nippon Roche Inc. and Sumitomo Pharmaceutical Inc. or purchased from GIBCO BRL (Gaitherburg, MD) and 8-bromoadenosine 3',5'-cyclic monophosphate (8Br-cAMP) was purchased from Sigma Chemicals (St Louis, MO). For cell cycle analyses, cell nuclei were prepared and stained with propidium iodide using CycleTEST (Becton Dickinson, San Jose, CA) according to the manufacturer's instructions. DNA ¯uorescences were measured with a FACScan ¯ow cytometer (Becton Dickinson, San Jose, CA), and the percentages of cells within G1, S and G2/M phases were determined (Matsushime et al., 1991). Antibodies The mouse monoclonal antibody D1-72-13G to mouse D1 and polyclonal rabbit antibodies to mouse cdk7 or MO15 were as described (Matsushime et al., 1994; Matsuoka et al., 1994). The polyvalent rabbit antisera to the C-terminal peptide of the mouse cdk4 (C-22), the C-terminal peptide of the mouse cdk2 (M2), p19 (M-20) (to the epitope corresponding to amino acids 135 ± 154 of mouse p19), p21 (C-19) (to the epitope corresponding to amino acids 146 ± 164 of mouse p21) and p27 (N-20) (to the epitope corresponding to amino acids 2 ± 21 of human p27 which can also recognize mouse p27), were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). The latter three antibodies recognized mouse p19, p21 or p27 and did not cross-react with each other. The polyclonal antibodies to cyclin E (the antibody to the epitope corresponding to a peptide, GVLTPPHSSKKQSSEQETE, mapping at the carbocyxterminus of rat cyclin E which can recognize mouse cyclin E), the monoclonal antibodies to human cyclin E (HE111) which recognized only human cyclin E

and were used for human cyclin E-associated kinase assays, and the polyclonal anti-cdc25A antibodies were also from Santa Cruz Inc. Protein analyses BAC1.2F5A cells (56106) were lysed for 1 h at 48C with 1 ml of EBC bu€er (50 mM Tris-HCl [pH 8.0], 120 mM NaCl, 0.5% NP-40, 1 mM EDTA and 1 mM dithiothreitol [DTT] or Tween 20 lysis bu€er (50 mM HEPES [pH 7.5], 150 mM NaCl, 1 mM EDTA, 2.5 mM EGTA, 1 mM DTT, 0.1% Tween 20) containing 5 mg/ml of aprotinin, 0.2 mM phenylmethylsulfonyl ¯uoride (PMSF), 10 mM b-glycerophosphate, 2.5 mg/ml leupeptin, 2.5 mM EGTA and 0.1 mM sodium orthovanadate (protease and phosphatase inhibitors from Sigma). Cleared cell lysates (100 ± 400 mg) were immunoprecipitated with saturated amounts of antibodies, separated on denaturing gels, transferred onto nitrocellulose ®lters, and blotted with indicated antibodies. Detection was with the ECL system (Amersham, Arlington Heights, Ill). For direct immunoblotting, cells were lysed in NP-40 lysis bu€er (20 mM Tris-HCl [pH 7.5], 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate) containing protease and phosphatase inhibitors. Aliquots (50 ± 100 mg) of clear lysates were loaded in each lane. Densitometric estimation of chemiluminescence-labeled protein amounts was performed with Quantity One software from PDI Inc. (Huntington, NY). Several ®lms with di€erent exposures were prepared for comparison within the linear correlation range of protein amount and signal strength. Immune complex kinase assays Kinase assays were performed as described previously (Matsushime et al., 1994). Brie¯y, cells were suspended at 46106/ml in Tween-20 lysis bu€er containing protease and phosphatase inhibitors, and lysed by sonication. Cyclin D1-, cdk4-, cyclin E- and cdk2-associated kinase activities were immunoprecipitated from cleared cell lysates (100 ± 500 mg) with protein A-Sepharose beads precoated with saturated amounts of antibodies to cyclin D1, cdk4, cyclin E and cdk2, and assayed for pRb kinase activities or histone H1. The beads were washed three times with 0.5 ml Tween 20 lysis bu€er and three times with kinase bu€er (50 mM HEPES [pH 7.5}, 10 mM MgCl2, 1 mM DTT), suspended and incubated for 1 h at 308C in 20 ml of the kinase bu€er containing 1 mg glutathione S-transferasepRb fusion protein or 1 mg histone H1 as substrates in association with protease and phosphatase inhibitors, supplemented with 25 mM ATP, 10 mCi of [r-32P]ATP (6000 Ci/mmol, New England Nuclear). For preparation of pRb substrates, pGEX-Rb-C (C-terminal polypeptide of pRb) or pGEX-Rb-Large Pocket (379 ± 928) was used. The precise isolation procedures were as described (Matsushime et al., 1994). Relatively higher background signals were observed with GST-Rb-C polypeptides as substrates than with GST-Rb-Large Pocket. The signal strength increased in parallel with the amounts of cell lysates (100 ± 500 mg) under this assay condition. 32P radioisotope counts of phosphorylated pRb were monitored with Bioimage Analyzer BAS-2000II (Fuji Photo Film Co, Ltd, Shizuoka, Japan). Kinase inhibition assays Immune complex cdk2- or cdk4-associated kinase assays were done after normal cell lysates (200 mg or 400 mg) were mixed with lysates containing cdk inhibitors and incubated for 30 ± 60 min at 238C. The lysis bu€ers were H1 kinase bu€er (80 mM Na-b-glycerophosphate, pH 7.3-15 mM MgCl2-20 mM EGTA {ethylene glycol-bis(b-aminoethyl ether)-N,N,N',N'-tetraacetic acid}-5 mM DTT) containing

Interferon a-induced G1 arrest M Matsuoka et al

protease and phosphatase inhibitors. The immunoprecipitates were washed three times with Tween 20 lysis bu€ers and three times with H1 kinase bu€ers before the kinase reaction was performed with the substrates of bacteriallyproduced pRb proteins or histone H1. Preparation of baculovirally-produced proteins Cyclin E, cyclin D2 and cdk4 were prepared by the method of Matsushime et al. (1992). Brie¯y, 60 mm plates of con¯uent TN5 cells were infected with the appropriate baculovirus. The cells were collected 48 ± 72 h after virus infection and lysed in H1 kinase bu€er for 1 h at 48C. The cleared lysates were directly used for cdk2 activation assays. Cdk2 activation assays Endogenous cdk2 activation in the cell lysates by exogenously-added cyclin E was as described by Polyak K et al. (1994a). Brie¯y, indicated amounts of baculovirally-generated human cyclin E were mixed with 150 mg of cell lysate in the H1 kinase bu€er containing protease and phosphatase inhibitors and incubated for 30 min at 238C before the reaction was immunoprecipitated with the cyclin E monoclonal antibody. The immunoprecipitates were washed and assayed for pRb or histone H1 kinase activities as described in the immune complex kinase assays. In some experiments, indicated amounts of baculovirally-synthesized cyclin D2, cdk4 or cyclin D2/ cdk4 complexes were added to the mixed lysates.

Cdk2 CAK assays Cells were suspended at 46106/ml in H1 kinase bu€er containing protease and phosphatase inhibitors and sonicated at 48C. The cleared supernatants were precipitated for 2 h at 48C with the antibodies to mouse cdk7 and the immunoprecipitates were washed three times with Tween 20 lysis bu€er and three times with H1 kinase bu€er. The beads were suspended in 20 ml of H1 kinase bu€er containing 5 mg of bacterially-produced His-tagged cdk2 as the CAK substrates supplemented with 50 mM ATP and 10 mCi [g-32P]ATP, and then incubated at 308C for 60 min. His-tagged cdk2 protein was isolated as described (Matsuoka et al., 1994). Northern blot analysis Total RNA was extracted from cells using the AGPC method. Northern blot hybridization was performed as described (Matsushime et al., 1991).

Acknowledgements We thank Sumitomo Pharmaceutical Inc. and Nippon Roche Inc. for providing mouse interferon-a, Drs Charles J Sherr and Hitoshi Matsushime for the baculoviruses and antibodies and Dr Jun-ya Kato for critical advice on the manuscript. We are also grateful to Drs Satoshi Takahashi and Arinobu Tojo for various supports. Daniel Murozek helped us to correct English.

References Chan FKM, Zhang J, Chen L, Shapiro DN and Winoto A. (1995). Mol. Cell. Biol., 15, 2682 ± 2688. Chin YE, Kiatgawa M, Su WC, You ZH, Iwamoto Y and Fu XY. (1996). Science, 272, 719 ± 722. Darnell JE, Kerr JM and Stark GR. (1994). Science, 264, 1411 ± 1420. Dulic V, Lees E and Reed SI. (1992). Science, 257, 1958 ± 1961. Dulic V, Kaufman WK, Wilson SJ, Tlsty TD, Lees E, Harper JW, Elledge SJ and Reed SI. (1994). Cell, 76, 1013 ± 1023. El-Deiry WS, Tokino T, Velculescu VE, Levy DB, Parsons R, Trent JM, Lin D, Mercer E, Kinzler KW and Vogelstein B. (1993). Cell, 75, 817 ± 825. Ewen ME, Sluss HK, Whitehouse LL and Livingston DM. (1993). Cell, 74, 1009 ± 1020. Fisher RP and Morgan DO. (1994). Cell, 78, 713 ± 724. Gorospe M, Liu Y, Xu Q, Chrest FJ and Holbrook NJ. (1996). Mol. Cell. Biol., 16, 762 ± 770. Gu Y, Rosenblatt J and Morgan DO. (1992). EMBO J., 15, 3995 ± 4005. Guan K, Jenkins CW, Li Y, Nichols MA, Wu X, O'Keete CL, Matera AG and Xiong Y. (1994). Genes & Dev., 8, 2939 ± 2952. Hall M and Peters G. (1996). Adv. Cancer Res., 68, 67 ± 108. Hannon GJ and Beach D. (1994). Nature, 371, 257 ± 261. Hara E, Smith R, Parry D, Tahara H, Stone S and Peters G. (1996). Mol. Cell. Biol., 16, 859 ± 867. Harper JW, Adami GR, Wei N, Keymarsi K and Elledge SJ. (1993). Cell, 75, 805 ± 816. Hatakeyama M, Brill JA, Fink GR and Weinberg RA. (1994). Genes & Dev., 8, 1759 ± 1771. Hobeika AC, Subramaniam PS and Johnson HM. (1997). Oncogene, 14, 1165 ± 1170. Hirai H, Roussel MF, Kato JY, Ashmun RA and Sherr CJ. (1995). Mol. Cell. Biol., 15, 2672 ± 2681. Ihle JN and Kerr IM. (1995). Trends Genet., 11, 69 ± 74. Jinno S, Suto K, Nagata A, Igarashi M, Kanaoka Y, Nojima H and Okayama H. (1994). EMBO J., 13, 1549 ± 1556.

Kamb A, Gruis NA, Weaver-Feldhaus J, Liu Q, Harshman K, Tavtigian SV, Stockert E, Day III RS, Johnson BE and Skolnick MH. (1994). Science, 264, 436 ± 440. Kato JY and Sherr CJ. (1993). Proc. Natl. Acad. Sci. USA, 90, 11513 ± 11517. Kato JY, Matsuoka M, Strom DK and Sherr CJ. (1994a). Mol. Cell. Biol., 14, 2713 ± 2721. Kato JY, Matsuoka M, Polyak K, Massague J and Sherr CJ. (1994b). Cell, 79, 487 ± 496. Ko€ A, Cross F, Fisher A, Schumacher J, Leguellec K, Phillippe M and Roberts JM. (1991). Cell, 66, 1217 ± 1228. Ko€ A, Giordano A, Desai D, Yamashita K, Harper JW, Elledge SJ, Nishimoto T, Morgan DO, Franza BR and Roberts JM. (1992). Science, 257, 1689 ± 1694. Ko€ A, Ohtsuki M, Polyak K, Roberts JM and Massague J. (1993). Science, 260, 536 ± 539. Lavarone A and Massague J. (1997). Nature, 387, 417 ± 422. Lee MH, Reynidottir I and Massague J. (1995). Genes & Dev., 9, 639 ± 649. Lew DJ, Dulic V and Reed SI. (1991). Cell, 66, 1197 ± 1206. Liu M, Lee MH, Cohen M, Bommakanti M and Freedman LP. (1996). Genes & Dev., 10, 142 ± 153. Li Y, Jenkins CW, Nichols MA and Xiong Y. (1994). Oncogene, 9, 2261 ± 2268. Lloyd AC, Obermuller F, Staddon S, Barth CF, McMahon M and Land H. (1997). Gene & Dev., 11, 663 ± 677. Matsuoka M, Kato JY, Fisher RP, Morgan DO and Sherr CJ. (1994). Mol. Cell. Biol., 14, 7265 ± 7275. Matsuoka S, Edwards M, Bai C, Parker S, Zhang P, Baldini A, Harper JW and Elledge SJ. (1995). Genes & Dev., 9, 650 ± 662. Matsushime H, Roussel MF, Ashmun RA and Sherr CJ. (1991). Cell, 65, 701 ± 713. Matsushime H, Ewen ME, Strom DK, Kato JY, Hanks SK, Roussel MF and Sherr CJ. (1992). Cell, 71, 323 ± 334. Matsushime H, Quelle DE, Shurtle€ SA, Shibuya M, Sherr CJ and Kato JY. (1994). Mol. Cell. Biol., 14, 2066 ± 2076.

2085

Interferon a-induced G1 arrest M Matsuoka et al

2086

Meyerson M and Harlow E. (1994). Mol. Cell. Biol., 14, 2077 ± 2086. Morgan DO. (1995). Nature, 374, 131 ± 134. Nobori T, Miura K, Wu DJ, Lois A, Takabayashi K and Carson DA. (1994). Nature, 368, 753 ± 756. Noda A, Ning Y, Venable SF, Pereira-Smith OM and Smith JR. (1994). Exp. Cell. Res., 211, 90 ± 98. Nourse J, Firpo E, Flanagan WM, Coats S, Polyak K, Lee MH, Massague J, Crabtree J and Roberts JM. (1994). Nature, 368, 753 ± 756. Ohtsubo M and Roberts JM. (1993). Science, 259, 1908 ± 1912. Parker SB, Eichele G, Zhang P, Rawls A, Sands AT, Bradley A, Olson EN, Harper JW and Elledge SJ. (1995). Science, 267, 1024 ± 1027. Pestka S, Langer JA, Zoon KC and Samuel CE. (1987). Annu. Rev. Biochem., 56, 72. Polyak K, Kato JY, Solomon MJ, Sherr CJ, Massague J, Roberts JM and Ko€ A. (1994a). Genes & Dev., 8, 9 ± 22. Polyak K, Lee M, Erdjument-Bromage M, Ko€ A, Roberts JM, Tempst P and Massague J. (1994b). Cell, 78, 59 ± 66. Poon RYC, Adamczewski JP, Hunt T and Shuttleworth J. (1993). EMBO J., 12, 3123 ± 3132. Quelle DE, Ashmun RA, Shurtle€ SA, Kato JY, Bar-Sagi D, Roussel MF and Sherr CJ. (1993). Genes & Dev., 7, 1559 ± 1571. Quelle DE, Ashmun RA, Hannon GJ, Rehberger PA, Trono D, Richter KH, Walker C, Beach D, Sherr CJ and Serrano M. (1995a). Oncogene, 11, 635 ± 645.

Quelle DE, Zindy F, Ashmun RA and Sherr CJ. (1995b). Cell, 83, 993 ± 1000. Resnitzky D and Reed SI. (1995). Mol. Cell. Biol., 15, 3453 ± 3469. Resnitzky D, Gossen M, Bujard H and Reed SI. (1994). Mol. Cell. Biol., 14, 1669 ± 1679. Reynisdottir I, Polyak K, Iavarone A and Massague J. (1995). Genes & Dev., 9, 1831 ± 1845. Reynisdottir I and Massague J. (1997). Genes & Dev., 11, 492 ± 503. Serrano M, Hannon GJ and Beach D. (1993). Nature, 366, 704 ± 707. Sherr CJ. (1993). Cell, 73, 1059 ± 1065. Sherr CJ and Roberts JM. (1995). Genes & Dev., 9, 1149 ± 1163. Terada Y, Tatsuka M, Jinno S and Okayama H. (1995). Nature, 376, 358 ± 362. Tiefenbrun N, Melamed D, Levy N, Resnitzky D, Ho€man I, Reed SI and Kimchi A. (1996). Mol. Cell. Biol., 16, 3934 ± 3944. Toyoshima H and Hunter T. (1994). Cell, 78, 67 ± 74. Tushinski RJ and Stanley ER. (1985). J. Cell. Physiol., 122, 221 ± 228. van den Heuvel S and Harlow E. (1993). Science, 262, 2050 ± 2054. Xiong Y, Hannon GJ, Zhang H, Casso D, Kobayashi D and Beach D. (1993). Nature, 366, 701 ± 704. Xu J, Rockow S, Kim S, Xing W and Li W. (1995). Mol. Cell. Biol., 14, 8018 ± 8024.